**4. Discussion**

It is now known that intestinal microbiota is one of the main elements capable of influencing immunity, health status, susceptibility to various diseases including chronic and autoimmune inflammatory diseases [13].

In literature, it has been reported that 25% fewer bacterial genes are detected in faecal samples of patients with IBD compared to the control groups. Furthermore, this reduction in diversity has been shown to occur early in the course of CD in a paediatric population, suggesting that dysbiosis may not only be an effect of CD, but also can contribute to the pathogenesis [14]. Further studies have shown that patients with IBD have fewer bacteria with anti-inflammatory properties and more bacteria with pro-inflammatory properties. Joossens et al. have identified stool samples containing microbiota from patients with CD with a reduced abundance of *Faecalibacterium prausnitzii*, *Bifidobacterium adolescentis* and *Dialister invisus* and a greater abundance of *Ruminococcus gnavus,* a potentially inflammatory bacterium [15]. The decrease in both biodiversity and in phyla *Bacteroidetes* and *Firmicutes* was observed in faecal and bioptic samples of patients with CD. Furthermore, many kinds of potentially protective bacteria, such as *Bacteroides*, *Eubacterium* and *Lactobacillus*, were significantly reduced in patients with active or inactive CD. *Roseburia*, a genus producing butyrate and *Phascolarctobacterium faecium* that produces propionate, have been found to be significantly reduced in patients with CD [16]. A study analysed faecal samples from a prospective cohort of patients with paediatric CD that underwent anti-TNF therapy. Regarding the dynamics of the microbiome (including viroma and micoma) with respect to therapy and diet, the dysbiosis decreased in concomitance with the reduction of the intestinal inflammation [17]. To characterise the intestinal microbiota associated with paediatric CD, Wang et al. recruited 11 children diagnosed with CD and healthy control subjects. A total of 32 samples of patients with CD were included: eight at baseline (before treatment with infliximab) and 24 at various times during therapy. Analysis of alpha diversity revealed that both wealth and diversity were lower in paediatric patients with CD before infliximab therapy compared to healthy controls. In particular, in the pre-infliximab samples a lower relative abundance of *Bacteroidetes* and a greater abundance of *Proteobacteria* were observed in patients compared to controls. After treatment, both the richness and the diversity of the intestinal microbiota improved in patients with paediatric CD. The community of bacteria in the post-infliximab samples was more similar to the control group, suggesting that the diversity between CD cases and healthy controls was reduced after treatment [18].

Changes in the composition of the faecal microbial community could therefore prove useful as biomarkers, in particular for monitoring disease activity, assessing the response to treatments [19] and as predictor of response to therapy [20,21].

In our study, we examined the relative percentage abundances of the four main bacterial phyla, namely *Firmicutes*, *Proteobacteria*, *Actinobacteria* and *Bacteroidetes*, of the family *Lachnospiraceae* and of the species *Bifidobacterium adolescentis*, *Faecalibacterium prausnitzii*, *Bacteroides Ovatus, Escherichia coli* and *Ruminococcus gnavus*. We focused on these taxa because each of them seems to have an interesting role in the pathophysiology of IBD: the *Lachnospiraceae* family (including several genera of *Clostridia* cluster XIVa, XIVb, IV and *Faecalibacterium prausnitzii*) is composed mainly of anti-inflammatory butyrogenic species and is reduced in patients with IBD, increasing proportionally to the remission of the disease [22]. *Ruminococcus gnavus* is a mucolytic bacterium found increased in IBD compared to healthy controls and is considered a possible biomarker of mucosal damage [19]. *Bifidobacteria* play a positive role in preserving intestinal barrier functions [23] and in the production of short-chain fatty acids (SCFA) [24]; of note, the analysis of the faecal microbiome of patients with IBD has shown an attenuation of *Bifidobacterium adolescentis* [25]. High antibody titres have been found targeting the antigens of *Bacteroides ovatus* [26], a bacterium that appears to be involved in the pathogenesis of IBD [27].

We assessed whether the taxa examined between the first faecal sampling and the second after six months of adalimumab therapy showed changes in terms of percentage abundance. It is interesting to note the course of the phylum *Proteobacteria* and of the family *Lachnospiraceae*. The former decreased significantly (*p* = 0.038), from 15.7 ± 3.5% to 10.3 ± 3.4%, while *Lachnospiraceae*increased from 18.2 ± 2.6% to 23.6 ± 2.2% (*p* = 0.100).

Bacterial concentrations before starting adalimumab therapy were considered in relation to achievement of therapeutic response. Although a predictive value of Firmicutes on response to therapy has been highlighted in the literature [28], and in particular of anti-inflammatory bacteria such as *Faecalibacterium prausnitzii* [20,28–31], this trend was not found in our study. Responder and non-responder patients had non-significant concentration differences of all taxa.

Then, we compared the trend of bacterial populations between T0 and T1 in those who responded versus those in whom the therapy failed. We found interesting modifications of both the phylum *Proteobacteria* and the family *Lachnospiraceae*: in those who responded to the therapy, the former decreased from T0 (15.8 ± 4.6%) to T1 (6.8 ± 3.1%) in a significant manner (*p* = 0.049). In those who did not respond to therapy, the trend was T0 = 15.6 ± 5.7%, T1 = 16.8 ± 7.6% (*p* = 0.890). With regards to the bacteria belonging to the *Lachnospiraceae* family, they increased more in responders (from 17.8 ± 3.3% to 25.4 ± 3.2%, *p* = 0.100) compared to those who did not respond (T0 = 18.8 ± 4.8%, T1 = 20.4 ± 1.8%, *p* = 0.730). With regards to the bacteria belonging to Proteobacteria phylum, *Escherichia coli* decreased from 11.4% to 4.3% (*p* = 0.078) in responders, while it remained substantially stable in those who did not respond (from 11.4% to 13.1, *p* = 0.81). Considering the trend of the intestinal microbiota during biologic therapy and the CRP values at the sixth month, there was an increase in the *Lachnospiraceae* family from T0 (16.6 ± 11%) to T1 (23.9 ± 9.6%) in patients who showed a normalization of CRP (significant: *p* = 0.049), while in those with persistent high CRP, it remained stable. The increasing trend of phylum Firmicutes and Lachnospiraceae family in patients with normalization of CRP is coherent (from 43.7–48.4% and from 16.6–23.9%, respectively): our explanation of the fact that in Lachnospiraceae family this trend is more evident is that, probably, Lachnospiraceae family, among the families belonging to phylum Firmicutes, is a species more represented in an "eubiotic" microbiota. The decrease of *Proteobacteria* and the increase of *Lachnospiraceae* is consistent with the hypothesis that adalimumab therapy, by decreasing inflammation, tends to restore the intestinal eubiosis [8,18,32].

The higher prevalence of the phylum Bacteroidetes in patients with mild or moderate endoscopic activity and the higher prevalence of the phylum Proteobacteria in patients with severe endoscopic activity confirm the potential role as protective bacteria of the former and as bacteria correlated to the inflammation of the latter.

Some limitations of our study must be discussed. The sample size of our population is not very large, although the prospective design contributes to reducing the possible biases. In all patients, diagnosis, treatment and follow-up of CD followed International Guidelines [33]. Another limit is that we focused only on some components of the human intestinal microbiota (according to literature data), even though viroma and micoma should add precious information on this topic.
